Investigation of Cook’s Distance Based on Robust Estimators of Variance

25219

In regression Cook’s Distance is more useful technique than the other methods due to its simple application about detect influential observations. Since residual sum of squares (SSRes) increases, σ^2 known as least likelihood estimate of variance of errors σ² has great values when data contains influential observations. In this case Cook’s Distance values calculated separately for each observation could be affected negatively from great values of the least likelihood estimate of variance. In that case of situations Cook and Weisberg (1982) proposed using robust estimations of variance for the Cook’s Distance. In this paper, Cook’s Distance has been investigated on robust estimators of σ².

8

Cooks Distance , Robust Estimators of Variance , Simulation.

Journal Of Natural an‎d Applied Sciences

14

Journal Of Natural an‎d Applied Sciences